Proceedings of Applied International Business Conference 2008
TESTING NONLINEAR CONVERGENCE IN MALAYSIA, 1965-2003
Muzafar Shah Habibullah a, ψ, A.M. Dayang-Affizza b, Venus Khim-Sen Liew c and Kian-Ping Lim c
b
a
Universiti Putra Malaysia, Malaysia
Universiti Malaysia Sarawak, Malaysia
c
Universiti Malaysia Sabah, Malaysia
Abstract
The purpose of the present paper is to examine income convergence in Malaysia by using the nonlinear
unit root test due to Kapetanios et al. (KSS, 2003) and extended by Chong et al. (CHLL, 2008) to
permit the test of long-run convergence and catching-up hypotheses. We apply the KSS-CHLL
nonlinear unit root for the test of nonlinear convergence between thirteen states with respect to Wilayah
Persekutuan (the riches state) of Malaysia for the period 1965 to 2003. Generally, our results suggest
that out of the thirteen states, only Kedah, Negeri Sembilan, Perak, Perlis and Selangor support the
long-run convergence hypothesis while Johor, Kelantan, Melaka, Pahang and Penang suggest catchingup. Lastly, Sabah, Sarawak and Terengganu indicate income divergence from Wilayah Persekutuan.
Keywords: Income convergence; Income divergence; Long-run convergence; Catching-up; Malaysia;
Nonlinear.
JEL Classification Codes: C32; F43; O40.
1. Introduction
During the last forty years, the government of Malaysia, through a series of five-year development
plans has implemented various policies and programmes aimed at alleviating regional disparities and
underdevelopment. The purpose of the regional development plans was to mainly focus on reducing the
large imbalances in social and economic development among states in the country. To correct these
imbalances is important because according to Hill (2002), regional economic disparities hamper
economic growth and that countries with a relatively even spatial distribution of income are likely to
grow faster. However, it was only during the Second Malaysia Plan 1971-1975 (Government of
Malaysia, 1971), that the government established the State Planning Units to enable the individual
states to identify and formulate projects and to coordinate development activities at their State level
(Tengku-Hadi, 1996).
On 4th November 2006, the Malaysian government has launched the most ambitious development
project in the region – the South Johor Economic Region (SJER) now known as the Iskandar
Development Region (IDR) which boasted some 800,000 jobs by the year 2020; comprises of an area
nearly three times the size of Singapore with passport free zone for foreigners; and expected to generate
an average rate of growth of eight percent for Johor. This day marked the serious effort of the
Malaysian government regional development plans laid out in the Ninth Malaysia Plan to reduce
regional imbalances and income disparity among states. In the Ninth Malaysia Plan 2006-2010
(Government of Malaysia, 2006: p. 363), there are five main thrusts for balanced regional development.
These includes: (1) accelerating development in lesser developed states through improving
infrastructure, social facilities amenities in the rural areas; (2) improving the quality of life in rural and
urban areas; (3) establishing new regional development authorities (RDAs) in Sabah and Sarawak; (4)
enhance higher economic growth through developing growth centres and growth corridors transcending
state boundaries; and (5) enhancing development of border states through ASEAN sub-regional
development cooperation in IMT-GT, BIMP-EAGA and JDS. 1 The subsequent so-called corridor
development projects include the Northern Corridor Economic Region (NCER), Eastern Corridor
Economic Region (ECER), Sabah Development Corridor (SDC) and Sarawak Corridor of Renewable
Energy (SCORE) which was launched recently.
ψ
Corresponding author. Muzafar Shah Habibullah. Faculty of Economics and Management, Universiti
Putra Malaysia, 43400 Serdang, Selangor, Malaysia. Email: [email protected]
1
IMT-GT denotes the Indonesia-Malaysia-Thailand Growth Triangle. BIMP-EAGA denotes the
Brunei Darussalam-Indonesia-Malaysia-Philippines East ASEAN Growth Area. JDS denotes the Joint
Development Strategy for Border Areas.
Proceedings of Applied International Business Conference 2008
The purpose of the present study is to determine whether the effort of the Malaysian government
through the various regional development plans for the past four decade has been successful in
narrowing the regional income gaps. In other words, in economic terms, we are questioning whether
the states in Malaysia have been converging, diverging or catching up. In this study, we used annual
data for the period 1965 to 2003, and based on the stochastic convergence definition suggested by
Bernard and Durlauf (1995) and using the nonlinear unit root test of Kapetanios et al. (KSS, 2003) and
extended by Chong et al. (CHLL, 2008) we be able to test the long-run convergence and catching-up
hypotheses.
The paper is organized as follow. In the next section we provide some stylized facts on the performance
of the fourteen states in Malaysia. In section 3, we provide some related literature on the issue of
nonlinear growth convergence. In section 4, we discuss the nonlinear unit root procedure employed in
the study. In section 5, we discuss the empirical results and the last section contains our conclusion.
2. Regional income disparity in Malaysia: some stylized facts
Economic convergence refers to a process in which national economies display increasing similarities
in the patterns of their performance. Convergence would point to the existence of market forces, which
will eventually lead to similar living standards across states. In the case of divergence (or widening
gaps) between poor and rich states, there could be a need for economic policy measures to stimulate a
catch-up process. The catching-up hypothesis suggests that the poorer states with low initial income
and productivity will tend to grow more rapidly by copying the technology from the leader state, say by
replacing existing older capital stock with more modern equipment, implying that capital investment is
necessary to import the more advanced technology embodied in new equipment (Lim and McAleer,
2002). One good example of transferring foreign technology and knowledge to the host country is
through foreign direct investment.
Whether poor economies tend to converge towards rich ones or else to diverge over time is an issue
that has attracted the attention of policy-makers and academics alike for some decades. Economic
convergence or divergence is a topic of considerable interest and debate, not only for validating or
otherwise the two leading and competing growth models (the neoclassical and the endogenous growth
approaches) but also for its policy-oriented implications. In Malaysia, the issue of economic
convergence is also much debated. Despite the various Malaysia Development Plans for the past four
decades, regional income disparity between states remains.
Table 1 and Table 2 show some interesting observations on the performance of the fourteen states in
Malaysia for the period 1970-2000. In the year 1970, five states- Negeri Sembilan, Perak, Selangor,
Sabah and Wilayah Persekutuan registered real GDP per capita that is above the national average.
However, in the year 2000, Melaka, Penang, Selangor, Terengganu and Wilayah Persekutuan has been
acting as the engine of growth, contributing to real GDP per capita that is above the national average.
Take for example the state of Sabah. In the year 2000, Sabah has been lagging behind the national
average by 35 percent of real GDP per capita. In terms of her ranking, in 1970, Sabah ranked third after
Wilayah Persekutuan and Selangor. However in 2000, Sabah ranked twelve followed by Kedah (13th)
and Kelantan (14th). The statistics suggest that in 2000 Sabah is the third poorest state in Malaysia,
despite her high ranking as the third richest states in 1970. The poor performance of the Sabah
economy has been recognized by the government of Sabah in the Outline Perspective Plan Sabah
(1995). They revealed the following facts (i) The State’s economy has been growing out of tandem
with the national economy, (ii) The growth of the States’s economy has been very erratic, (iii) The
economy is still dominated by the primary sector, (iv) Unemployment remains persistently high, (v)
The investment ratio is low by national standard coupled with a probable massive outflow of funds
from Sabah, (vii) Rapidly depleting timber and petroleum resources, (viii) Limited sources of economic
growth, and (ix) Low value-adding economic activities. As for Kedah, it was ranked 11th in 1970, but
since 1980 the state of Kedah has been the second poorest state in the country. Kelantan, however,
remain the poorest of all the states in Malaysia for the last four decade. In summary, for four decades,
the lagging states include Kedah, Perak, Perlis, Negeri Sembilan, Selangor, Kelantan, Pahang, Sabah
and Sarawak, while the faster growing states include Penang, Melaka, Wilayah Persekutuan and
Terengganu.
573
Proceedings of Applied International Business Conference 2008
Table 1: Real GDP per capita, 1970-2000 (Malaysia=100)
States
1970
1980
1990
2000
Kedah
73
61
59
60
Perak
103
93
79
81
Perlis
72
60
66
66
Penang
96
113
118
143
Melaka
72
75
83
104
Negeri Sembilan
104
101
84
93
Selangor
148
156
142
124
Wilayah Persekutuan
176
197
191
205
Kelantan
44
60
38
42
Pahang
93
79
82
67
Terengganu
81
71
159
154
Johore
84
89
91
96
Sabah
118
101
85
65
Sarawak
92
80
88
90
Malaysia
100
100
100
100
Notes: Compiled and computed from various issues of the Five Year Malaysia Development Plans.
Table 2: Ranking by states according to real GDP per capita, 1970-2000
States
1970
1980
1990
2000
Kedah
11
13
13
13
Perak
5
9
11
9
Perlis
12
12
12
11
Penang
6
4
4
3
Melaka
13
10
9
5
Negeri Sembilan
4
5
8
7
Selangor
2
2
3
4
Wilayah Persekutuan
1
1
1
1
Kelantan
14
14
14
14
Pahang
7
6
10
10
Terengganu
10
3
2
2
Johore
9
8
5
6
Sabah
3
7
7
12
Sarawak
8
11
6
8
Notes: Compiled and computed from the issues of the Five Year Malaysia Development Plans.
3. Review of related literature
According to neoclassical growth models for closed economies (Solow, 1956), given similar
preferences and technology, the assumption of diminishing marginal product of capital will lead to
poor countries grow faster to catch-up with rich countries and hence results in absolute convergence
among countries. This happens as capital in a higher per capita income country which subject to
‘diminishing returns’, moves outward seeking opportunities in a country with a comparatively lower
per capita income (where new investments can be expected to benefit from a relative increase in rates
of output per unit of capital input, as each addition to capital stock generates enormous additions to
output when the capital stock is small). Thus, capital movement between countries serves as the
primary instrument driving economic convergence. Economic convergence is attained when
differences in rates of marginal returns to capital between countries is equal to zero. When such occurs
it is assumed that income per capita would also have equalized between countries.
In a time-series perspective according to Bernard and Durlauf (1995), the so-called stochastic
convergence asks whether permanent movements in one country’s per capita income are associated
with permanent movements in another countries’ income, that is, it examines, whether common
stochastic elements matter, and how persistent the differences among countries are. Thus, stochastic
convergence implies that income differences among countries cannot contain unit roots. In other words,
income per capita among countries is stationary. Empirical studies on testing stochastic convergence,
among others include Bernard (1991), Bernard and Durlauf (1995), Campbell and Mankiw (1989),
Cogley (1990), Greasly and Oxley (1997), St. Aubyn (1999), Cellini and Scorcu (2000) and Carlino
and Mills (1993). According to Bernard and Durlauf (1995), the notion of convergence in multivariate
574
Proceedings of Applied International Business Conference 2008
income is defined such that the long-term forecasts of income for all regions i = 1, . . . , n, are equal at a
fixed time t:
E ( y1,t + k − y i ,t + k I t ) = 0,
∀i > 1
(1)
where y i ,t + k is the logarithm of real per capita income for region i at time t + k , and I t is all the
information available at time t . Using the concepts of unit roots and cointegration, their convergence
test determines whether y1,t + k − y i ,t + k in Equation (1) is a zero mean stationary process in a
cointegration framework. Convergence in output for two regions, i and j , implies their income must
be cointegrated with cointegrating vector [1, − 1] . This concept of convergence has been criticized
because it is rather strict, as for the strong convergence to exist it is necessary that the long-run
expected value (forecast) of the per capita income differences between the two regions is equal to zero.
An alternative time-series definition of convergence, according to Bernard and Durlauf (1996) also
known as catch-up holds when the “behaviour of the income differences between two regions over a
fixed time interval and equates convergence with the tendency of the difference to narrow” (p. 165).
This definition can be written as
E ( y i ,T − y j ,T I t ) < ( y i ,0 − y j ,0 )
(2)
where 0 refers to the present and T to some year in the future. According to this definition, the
difference between the two time series should also be stationary, but now the time trend can be
deterministic. Once again, the only cointegration vector between the two regions can be [1, − 1] .
Following Bernard and Durlauf (1995), stochastic convergence occurs if relative log per capita GDP,
y it , follows a stationary process, where y iqt = log Yit − log Yqt , and Yit is the log of real per capita
GDP for country i , and Yqt is log of real per capita GDP of a benchmark country, and both series is
I (1) . Stochastic convergence is then tested by using the conventional augmented Dickey-Fuller (ADF)
regression of the following form
p
∆y iqt = α i + γt + β i y iqt −1 + ∑ θ ij ∆y iqt − j + ε iqt ,
j =1
t = 1,..., T
(3)
for i = 1,..., N countries, and j = 1,..., p ADF lags. In a time series framework, a distinction is made
between long-run convergence and convergence as catching-up (see Oxley and Greasley, 1995). The
statistical tests are interpreted as follows. First, if y iqt contains a unit root (i.e. β = 1 ), real GDP per
capita for states i and q diverge over time. Second, if
y iqt is stationary (i.e. no stochastic trend, or
β < 1 ) and (a) γ = 0 (i.e the absence of a deterministic trend) indicates long-run convergence between
states i and q ; (b) γ ≠ 0 indicates catching-up (or narrowing of output differences) between states
i and q .
However, according to Kapetanios et al. (2003) one important drawback of the ADF unit root test
procedures is that the power of the test is quite low in the presence of nonlinearities in the dynamic of
the variables and hence they might not be able to distinguish between unit root and nonlinear stationary
process. Potential failure to reject nonstationarity may be the result of linear unit root tests not being
very powerful when the true adjustment process is nonlinear (Gregoriou and Kontonikas, 2006). In fact
studies have found that some macroeconomic variables exhibit nonlinearities. For example in testing
for the purchasing power parity hypothesis, numerous studies have found that adjustment towards PPP
maybe nonlinear. One potential source arises from nonlinearities in international goods arbitrage
because of factors such as transportation costs and trade barriers causing a price gap among similar
goods traded in spatially separated markets. These costs and barriers are much higher in developing
countries than industrial countries, suggesting a strong case for nonlinear adjustment towards PPP in
these countries.
575
Proceedings of Applied International Business Conference 2008
In the growth literature, different theories suggest that economic growth is nonlinear. According to
Lewis (1956), Rostow (1960), Mas-Collel and Razin (1973), Murphy et al. (1989), and Galor and Weil
(2000), the growth path of an economy displays an initial phase of stagnation, followed by a take-off in
which growth rates are increasing and eventually reaches a regime of steady growth. These different
growth regimes associated to different levels of development, are generated by the structural
transformations faced by a growing economy. Peretto (1999) argues that a nonlinear growth process is
the result of the transition from growth generated by capital accumulation, subject to decreasing returns
to scale, to growth based on knowledge accumulation. Azariadis and Drazen (1990) and Durlauf and
Johnson (1995) reject the linear model commonly used to study cross-country growth behaviour in
favour of a multiple regime alternative in which different economies obey different linear models when
group together according to initial conditions. According to Azariadis and Drazen (1990) the
multiplicity is due to increasing social returns to scale in the accumulation of human capital.
Another reason is on growth focuses on different kinds of interactions which may take place among
economies. This literature devoted particular attention to technological spillovers (Parente and Prescott,
1994; Basu and Weil, 1998). These contributions analyze the conditions that allow a country, starting
its development process, to benefit from the knowledge accumulated by richer countries, and therefore
increase its growth rate. In this setting a nonlinear growth path could be the result of different adoption
speeds, when the speed increases as a country develop. Dobson et al. (2003) found that convergence is
not widespread, occurring among countries with very low and very high initial income levels. The
finding of nonlinearity lends credence to the idea that convergence clubs characterize the cross-country
growth process and that there is a clustering of countries in economic growth performance.
Furthermore, nonlinear economic growth can also be derived as a result of gradual reform strategy.
According to Lai (2006) China has entered a convergent growth path since 1978. The main driving
forces for the convergence is market-oriented reform and opening to the outside world. The main
mechanism through which market and opening drive convergence is essentially similar to the same
mechanism that works in the East Asian model (the flying geese pattern). This convergence is a
nonlinear one with serious ups and downs. The main reason is the fluctuation in reform and some
inappropriate development policies (such as government-led excessive investment), which is closely
associated with excessive state intervention in markets and enterprises.
On the other hand, Potter (1995) examines the nonlinear behaviour of U.S. GNP and found that the
univariate nonlinear model outperforms standard linear models. In fact, the nonlinear model suggests
that the post-1945 U.S. economy is significantly more stable than the pre-1945 U.S. economy. Liew
and Lim (2005), Liew and Ahmad (2007) and Chong et al. (2008) have investigated the issue of
nonlinear income convergence between countries. Using Kapetanios et al. (2003) nonlinear unit root
test, Liew and Lim (2005) found that Hong Kong, Taiwan and Singapore show convergence with Japan,
while China, Indonesia, Malaysia, Thailand and the Philippines show divergence. A study by Liew and
Ahmad (2007) on Japan, Hong Kong, Korea, Taiwan and Singapore, found that Hong Kong, Korea and
Singapore are catching-up while Taiwan has yet to catch-up with the Japan economy. Chong et al.
(2008) examine the long-run convergence and catching-up hypotheses between 15 OECD countries
relative to the U.S. Among others, their result suggests that Austria and the Netherlands exhibits longrun converging with the U.S., while Australia, Sweden, Switzerland and the U.K. are in the process of
catching-up.
4. Method of estimation
In this study we utilized the KSS nonlinear unit root test proposed by Kapetanios et al. (2003) to
determine the long-run converging and catching-up between the 14 states in Malaysia. The KSS test
analyses nonstationarity under the null hypothesis against the alternative of nonlinear but globally
stationary exponential smooth transition autoregressive (ESTAR) processes. The KSS test is based on
the following ESTAR specification
[
)]
(
2
∆~
y iqt = γ~
y iqt −1 1 − exp − θ~
y iqt
−1 + ε t
(4)
is the error term with zero mean and constant variance. Variable ~y iqt is the de-meaned or de2
yiqt
trend series of y iqt , and 1− exp −θ~
−1 is the exponential transition function adopted in the test to
y implies that H : θ = 0 (i.e.
present the nonlinear adjustment. The null hypothesis of a unit root in ~
where
εt
[
(
)]
iqt
576
0
Proceedings of Applied International Business Conference 2008
2
[1 − exp(− θ~
y iqt
−1 )] = 0 ), while the alternative of a nonlinear but globally stationary process requires that
H 1 : θ > 0, where θ effectively determines the speed of mean reversion.
In Equation (4) the parameter γ is not identified under the null and thus cannot be estimated directly
and test H 0 : θ = 0 . To circumvent this problem Kapetanios et al. suggest to reparametize Equation (4)
by computing a first-order Taylor series approximation to specification (4) to obtain the auxiliary
regression as follows
3
∆~
y iqt = δ~
y iqt
−1 + ω t
where
ωt
(5)
is a stochastic error term. To whiten the noise, the following augmented regression is
estimated
3
∆~
y iqt = δ~
y iqt
−1 +
k
∑ ϕ ∆~y
i
iqt −i
+ ωt
(6)
i =1
with k augmentations to correct for serially correlated errors. To test the null hypothesis of
nonstationarity H 0 : δ = 0 against the alternative H 1 : δ < 0 , the t-statistic (KSS statistic) tabulated by
Kapetanios et al. (2003) is used.
Our interest is to determine the long-run converging and catching-up and to distinguish between the
two hypotheses. Although the KSS nonlinear unit root test can detect nonlinear stationarity in the series,
however the KSS test using either Equation (5) or Equation (6) cannot distinguish between the two
concept (Liew and Ahmad, 2007; Chong et al., 2008). To circumvent this problem, Chong et al. (2008)
modified Equation (6) by including an additive intercept µ t and trend component [G(trend)] as follows
3
∆y iqt = µ t + φG (trend ) + δy iqt
−1 +
k
∑ ϕ ∆y
i
iqt −i
+ ξt
(7)
i =1
where
ξ t is the error term. From Equation (7), the absence of nonlinear unit root (δ
< 0) implies either
nonlinear catching-up, given the presence of deterministic trend (φ ≠ 0) , or nonlinear long-run
converging if deterministic trend is absent (φ = 0) . However, if y iqt contains a nonlinear unit root
(δ = 0) , the income between country
i and country q is said to diverge over time.
Description and Sources of Data
In this study the fourteen states in Malaysia are Johor, Kedah, Kelantan, Melaka, Negeri Sembilan,
Perak, Perlis, Pahang, Penang, Selangor, Sabah, Sarawak, Terengganu and Wilayah Persekutuan. Data
on states real gross domestic product were compiled from the various issues of the Five-Year Malaysia
Plans and the Mid-Term Review of the Malaysia Plans. For the year 1961 to 2003, twenty-four data
points were collected and for each data point were converted to the same base year, 2000=100. For the
remaining missing data point, we interpolate using regression with trend component (time, time square
and/or time cube) and current and one-year lagged Malaysia’s real GDP. In the analysis, all income
series were deflated by the respective states population to arrive at the real GDP per capita. The real
GDP per capital are transform into logarithm for the analysis throughout the study using data from
1965 to 2003.
577
Proceedings of Applied International Business Conference 2008
5. The empirical results
The estimated coefficients of estimating Equation (7) with linear trend are reported in Table 3. When
estimating Equation (7), we allow augmentation up to three years lag. The final estimated equations are
chosen by paring down the lag length until the last lag showing significance at the 10 percent level.
Results in Table 3 show the testing of nonlinear unit root on the income differential with respect to
Wilayah Persekutuan as the reference state.
Table 3: Results of KSS-CHLL test with constant and linear trend
φ
Lag δ
Remarks
Coefficient t-statistic
Coefficient t-statistic
Johor
1
-0.2883
-5.5095*** 0.00136
2.7892*
Catching-up
Kedah
1
-0.1108
-4.3139*** -0.00127
-1.2792
Convergence
Kelantan
3
-0.1724
-4.9740*** -0.01042
-3.9556*** Catching-up
Melaka
2
-0.1493
-4.1889*** 0.00535
4.6447***
Catching-up
Negeri Sembilan
2
-0.4346
-3.7972**
-0.00351
-2.0981
Convergence
Perak
1
-0.1821
-3.0153
-0.00151
-1.2230
Divergence
Perlis
1
-0.1419
-3.0658*
-0.00140
-1.1323
Convergence
Pahang
1
-0.1515
-1.8068
-0.00270
-1.1620
Divergence
Penang
0
-0.5971
-5.3403*** 0.00462
4.4643***
Catching-up
Selangor
0
-1.1961
-3.4353**
-0.00280
-1.8419
Convergence
Sabah
1
-0.1473
-2.7852
-0.00780
-2.9486*
Divergence
0.0322
Divergence
Sarawak
1
-0.2277
-2.5127
3.7 x 10-5
Terengganu
0
-0.1635
-2.5082
0.00517
2.4369
Divergence
Notes: Asterisks (***), (**), and (*) denote statistically significant at the 1%, 5% and 10% level
respectively. Critical values are referred to Chong et al. (2008), Table 1a for the t-statistic of δ ; while
Table 1b for the t-statistic of φ .
Series
The significance of the parameter δ indicates rejection of the null hypothesis of no income
convergence. In Table 3, unit root is found for Perak, Pahang, Sabah, Sarawak and Terengganu. This
implies that income of these five states diverge with respect to Wilayah Persekutuan. Income
convergence with Wilayah Persekutuan is found for Johor, Kedah, Kelantan, Melaka, Negeri Sembilan,
Perlis, Penang, and Selangor. The findings of income convergence permit us to test whether these states
attain long-run convergence or catching-up with respect to Wilayah Persekutuan.
The insignificant of the parameter φ will suggest long-run convergence or otherwise the catching-up
hypothesis. It is observed from Table 3 that the significance of the parameter φ is shown for Johor,
Kelantan, Melaka, and Penang. In the case of Kedah, Negeri Sembilan, Perlis, and Selangor support
long-run convergence with respect to the richer state, Wilayah Persekutuan.
On the other hand, Table 4 shows the results for nonlinear convergence including nonlinear trend in
Equation (7). Generally, the results are more overwhelming in the sense that the significance level
increases in some cases from 10 percent to 5 percent compare to the results in Table 3. For example,
for the state of Johor, φ is significantly different from zero at the 5 percent compare to 10 percent in
Table 3. This suggests that nonlinear trend is more representative of the Malaysian states income data.
More interesting are the results for the states of Perak and Pahang. Using linear trend, these two states
indicate divergence but when using nonlinear trend Perak indicates long-run convergence while Pahang
suggests catching-up. Overall, our results suggest that except for the states of Sabah, Sarawak and
Terengganu (showing divergence), long-run convergence with Wilayah Persekutuan are shown by
Kedah, Negeri Sembilan, Perak, Perlis, and Selangor. The states that are catching-up with Wilayah
Persekutuan include Johor, Kelantan, Melaka, Pahang and Penang.
578
Proceedings of Applied International Business Conference 2008
Table 4: Results of KSS-CHLL test with constant and nonlinear trend
φ
Lag δ
Remarks
Coefficient t-statistic
Coefficient t-statistic
Johor
1
-0.3111
-6.0744*** 2.91 x 10-5
3.0357**
Catching-up
Kedah
2
-0.1133
-5.3188*** -1.17 x 10-5
-1.1036
Convergence
-4.1947*** Catching-up
Kelantan
3
-0.1564
-5.3339*** -1.90 x 10-4
4.1883***
Catching-up
Melaka
1
-0.1626
-4.2073*** 9.41 x 10-5
-1.7939
Convergence
Negeri Sembilan
2
-0.3332
-4.1573*** -4.12 x 10-5
-1.0303
Convergence
Perak
1
-0.1521
-3.4519**
-1.89 x 10-5
-0.9153
Convergence
Perlis
1
-0.1266
-3.1180*
-2.03 x 10-5
-4.3352*** Catching-up
Pahang
0
-0.3882
-5.0286*** -1.92 x 10-4
4.4595***
Catching-up
Penang
0
-0.6263
-5.3950*** 9.78 x 10-5
-2.3478
Convergence
Selangor
0
-1.3467
-3.8555**
-7.30 x 10-5
-2.8627*
Divergence
Sabah
0
-0.2085
-2.9504
-2.05 x 10-4
0.3010
Divergence
Sarawak
1
-0.2225
-2.5399
6.89 x 10-6
1.1420
Divergence
Terengganu
3
-0.1091
-1.6621
5.06 x 10-5
Notes: Asterisks (***), (**), and (*) denote statistically significant at the 1%, 5% and 10% level
respectively. Critical values are referred to Chong et al. (2008), Table 1a for the t-statistic of δ ; while
Table 1b for the t-statistic of φ .
Series
6. Conclusion
The relative gap between the richest and poorest countries is a never ending story. Voluminous
research has been done to understand and explain disparity between (both across and within) countries
in both theoretical and empirically. According to Barro and Sala-i-Martin (1995), it is important to
identify the causes and nature of differences in levels and growth of income across countries (states or
regions or provinces) because even small differences in the growth rates, if accumulated over a long
period of time, may have substantial impact on the standard of living and also results in unnecessary
human suffering and an enormous squandering of human potential. In Malaysia, despite four and half
decades of development planning aiming to reduce the income disparity among states, the income
imbalances still persist.
In this study we investigate empirically whether the fourteen states in Malaysia exhibit long-run
income convergence by using the KSS-CHLL nonlinear unit root test to test for the present of
nonlinear convergence (versus divergence). Using the KSS-CHLL procedure we will be able to
distinguish between the long-run convergence and catching-up hypotheses. Using time-series data for
the period 1965 to 2003, and the richer state- Wilayah Persekutuan as the benchmark state, our results
suggest that out of the thirteen states, only Sabah, Sarawak and Terengganu indicate divergence from
Wilayah Persekutuan. On one hand, Kedah, Negeri Sembilan, Perak, Perlis, and Selangor suggest longrun convergence, and on the other hand, catching-up is shown by Johor, Kelantan, Melaka, Pahang and
Penang.
Generally, our results suggest that the nonlinear approach to convergence able to uncover economic
convergence among the states in Malaysia for the period under study. An important implication of this
study is that it appears that the Malaysian regional policies have an impact on the relative positions of
the fourteen states in terms of their share of Malaysia’s GDP. It is also important to recognise that
states income per capita increased in all of the fourteen states and that the extent of achieving regional
convergence might well have been lower in the absence of the regional policy, in particular, the various
Five-Year Malaysia Plan.
References
Azariadis, C. and Drazen, A. (1990) Threshold externalities in economic development. The Quarterly
Journal of Economics, 105, 2, 501-526.
Barro, R., and Sala-i-Martin, X. (1995) Economic Growth. New York: McGraw-Hill.
Basu, S. and D.N. Weil (1998) Appropriate technology and growth. The Quarterly Journal of
Economics, 113, 1025-1054.
Bernard, A.B., and Durlauf, S.N. (1995) Convergence in international output. Journal of Applied
Econometrics, 10, 97-108.
Bernard, A.B., and Durlauf, S.N. (1996) Interpreting tests of the convergence hypothesis. Journal of
Econometrics, 71, 161-173.
579
Proceedings of Applied International Business Conference 2008
Campbell, J.Y., and Mankiw, N.G. (1989) International evidence on the persistence of economic
fluctuations. Journal of Monetary Economics, 23, 319-333.
Carlino, G.A., and Mills, L.O. (1993) Are U.S. regional incomes converging? A time series analysis.
Journal of Monetary Economics, 32, 335-346.
Cellini, R., and Scorcu, A. (2000) Segmented stochastic convergence across the G-7 countries.
Empirical Economics, 25, 463-474.
Chong, T.T.L., Hinich, M.J., Liew, V.K.S. and Lim, K.P. (2008) Time series test of nonlinear
convergence and transitional dynamics. Economics Letters, 100, 337-339.
Cogley, T. (1990) International evidence on the size of the random walk in output. Journal of Political
Economy, 98, 501-518.
Dobson, S., Ramlogan, C., and Strobl, E. (2003) Cross-country growth and convergence: A semiparametric analysis. Unpublished Economics Discussion Papers No. 301, University of Otago,
New Zealand.
Durlauf, S.N. and Johnson, P.A. (1995) Multiple regimes and cross-country growth behaviour. Journal
of Applied Econometrics, 10, 365-384.
Galor, O. and D.N. Weil (2000) Population, technology and growth: from the Malthusian regime to the
demographic transition and beyond. American Economic Review, 90, 806-828.
Government of Malaysia. (1971) Second Malaysia Plan. Kuala Lumpur: Government Printers.
Government of Malaysia. (2006). Ninth Malaysia Plan. Kuala Lumpur: Government Printers.
Greasley, D., and Oxley, S. (1997) Time-series based tests of the convergence hypothesis: Some
positive results. Economics Letters, 56, 143-147.
Gregoriou, A. and Kontonikas, A. (2006) Inflation targeting and the stationarity of inflation: New
results from an ESTAR unit root test. Bulletin of Economic Research, 58, 4, 309-322.
Hill, H. (2002) Spatial disparities in developing East Asia: A survey. Asian-Pacific Economic
Literature, 16, 1, 10-35.
Kapetanios, G., Shin, Y., and Snell, A. (2003) Testing for a unit root in the nonlinear STAR framework.
Journal of Econometrics, 112, 359-379.
Lai, P (2006) China’s macroeconomic development: Stages and nonlinear convergence. China &
World Economy, 14, 15-29.
Lewis, A (1956) The Theory of Economic Growth. London: Allen and Unwin.
Liew, V.K.S. and Ahmad, Y. (2007) Long run convergence and catching up in income convergence:
Evidence from the four Asian dragons. Journal of International Economic Review, 1, 2, 121128.
Liew, V.K.S. and Lim, K.P. (2005) Income divergence? Evidence of non-linearity in the East Asian
economies. Economic Bulletin, 15, 1, 1-7.
Lim, L.K. and McAleer, M. (2002) Economic growth and technological catching up by Singapore to
the USA. Mathematics and Computers in Simulation, 59, 133-141.
Mas-Collel, A. and A. Razin (1973) A model of intersectoral migration and growth. Oxford Economic
Papers, 25, 72-79.
Murphy, K., A. Shleifer and R. Vishmy (1989) Industralization and the big push. Journal of Political
Economy, 97, 1003-1026.
Oxley, L. and Greasley, D. (1995) A time-series perspective on convergence: Australia, UK and USA
since 1870. The Economic Record, 71, 214, 259-270.
Parente, S.L. and E.C. Prescott (1994) Barriers to technology adoption and development. Journal of
Political Economy, 102, 298-321.
Peretto, P.F. (1999) Industrial development, technological change and long-run growth. Journal of
Development Economics, 60, 389-417.
Potter, S.M. (1995) A nonlinear approach to US GNP. Journal of Applied Econometrics, 10, 109-125.
Rostow, W.W. (1960) The Stages of Economic Growth. Oxford: Oxford University Press.
Solow, R. (1956) A contribution to the theory of economic growth. The Quarterly Journal of
Economics, 70, 65-94.
St. Aubyn, M. (1999) Convergence across industrialised countries (1890-1989): New results using time
series methods. Empirical Economics, 24, 23-44.
Tengku-Hadi, T.I. (1996) Langkawi: The Hub of the Indonesia, Malaysia, Thailand Growth Triangle.
Petaling Jaya: Pelanduk Publications.
580